Purely functional, embedded array programs are a good match for SIMD hardware, such as
GPUs. However, the naive compilation of such programs quickly leads to both code …

Commodity many-core hardware is now mainstream, but parallel programming models are
still lagging behind in efficiently utilizing the application parallelism. There are (at least) two …

Previous work has demonstrated that it is possible to generate efficient and highly parallel
code for multicore CPUs and GPUs from combinator-based array languages for a range of …

Regular array languages for high performance computing based on aggregate operations
provide a convenient parallel programming model, which enables the generation of efficient …

▶ NESL was designed for bulk-data processing on wide-vector machines (SIMD)▶
Potentially a good fit for GPU computation▶ First try [Bergstrom & Reppy'12] demonstrated …

A barrier to efficient array programming, for example in Python/NumPy, is that algorithms
written as pure array operations completely without loops, while most efficient on small input …

The paradigm of nested data parallelism (NDP) allows a variety of semi-regular computation
tasks to be mapped onto SIMD-style hardware, including GPUs and vector units. However …

Abstract Graphics Processing Units (GPUs) have hundreds to thousands of cores, promising
extremely high performance for data-parallel applications. However, most general-purpose …

In this thesis, we investigate streaming as a general solution to the space inefficiency
commonly found in functional data-parallel programming languages. The data-parallel …

With the mainstream emergence of many-core architectures, such as GPGPUs, massive
parallelism has become a focus area of industrial application development. However …